A check is inconclusive when the compared observable differs between repetitions, a run fails to load or a file changes during it. A disagreement on another observable stays flagged without voiding the comparison. Core log lines name the staged directories instead of the host paths.
core_probe.py loads a libretro core with no window, runs it for a
number of frames and reports the files it opened (strace), its log
lines, frame, audio and save-state hashes and its speed. bench repeats
a run after warm-up and gives median, mean, deviation and range; check
swaps one system file between runs and compares one observable.
--record writes a JSON record with the core build, frontend commit,
host and every input by hash, validated against
core-probe-record.schema.json. A busy host is refused, and a
repetition that disagrees with the first is flagged.
A project that tags without releases (eightyone, higan) was never compared,
and a core_version naming no tag (parallel-launcher 9.0.5 beside v9.0-6)
silenced the passed-release warning. The newest tag stands in when it is a
version number, and an untagged core_version is compared by number within
one scheme. --emulator is repeatable: a stored value kept only the last.
A profile pinned on its release tag read as clean forever: armsx1, armsx2
and x1-box sat on old tags while newer releases were out. --check-version
now names the latest release when it follows both the declared tag and
the pin, unless pin_frozen holds the pin on purpose.
A sitesources run that GitLab rate-limited dropped about twenty trees it
never read from website/src/data/source-paths.json. A tree key names a
pinned sha, so the last record's entry for an unread tree still holds and
is carried over, in sitesources and in profile_sync's refresh.
Refs that still anchor say nothing of what the code added around them:
mame2003_plus's zn.c gained two Taito G-NET BIOS sets while its cited
lines only shifted. --rebase-refs and --bump-commit now refuse a profile
whose cited files gained a file name or a hash literal since the pin, or
could not be read, until --accept-changed says the additions were read.
With this, --all --rebase-refs --fresh is safe to run as is.
citron's taken domain now resolves to an unrelated host that fails TLS,
so the pass reported an upstream error where the profile already states
the forge is gone. Any forge error now agrees with upstream_gone; a rate
limit still does not.
Each pin advance left website/src/data/source-paths.json behind and the
site's freshness test red until sitesources.py --fetch ran by hand.
profile_sync now rebuilds it after any run that wrote a profile, online
and not in a dry run.
The Apple II page named its start files with no visible way to get one: the download sat after the collection path, fourth of a card's facts. Each card now carries a Download button in its title, each start-line name links to its card, and a card held on another part page is reached by its permanent link.
--add-app, the result screen and a [y/N] question put the launcher of
muOS, KNULLI or ROCKNIX on the card the BIOS folder is on. The files come
from install/launchers/index.json at the same ref, each checked by size
and SHA-256 before the first is written, and the app event reports it.
Branch tips, tags and releases are cached for six hours, and the
EKA2L1, trzy/Supermodel and VICE heads read as behind their real HEAD.
--fresh asks the forges again, and is refused with --offline.
A recale on a profile whose pins are keyed by build mode said the prose
had to move first. It now says the pins are set by hand.
The README, the installer steps, the choice of what to install and platform detection still described the line prompts as what a terminal shows. They now name the full-screen interface first, with --plain for the questions, and say that no default folder is taken where detection was refused.
A platform list that declares one path for two files fails one of the
two checks on every install. The export report lists each declaration
with its hash and the emulators whose profile reads that file.
_MISSING_DATA_DIRS is module state: a test that called main() after one
whose data directories were not cached exited 1 with the earlier run's
notice. main() now clears it first.
The collector deduplicated a destination already carrying the extras
prefix at prefix/prefix/name while the builder wrote it at prefix/name.
DuckStation's standalone bios/<name> on Recalbox and PCSX2's scan then
held one path under two keys, and --region withdrew only one. The
second pass also no longer claims a path another core's extra holds
for a name only that extra covers.
A branch tip is cached for six hours. scummvm-standalone was pinned on
6f7b060d while the cache still held 77d80d7, its ancestor, so profile_sync
showed a HEAD behind the pin and a bump would have moved it backwards. A
tip that is a strict ancestor of the pin is now asked for again.
Every board set sits on its board page whatever folder holds it, a set a
reviewed machine owns joins that machine (neogeo.zip under SNK, the Atom
under Acorn, the MZ-2000 under Sharp), and a game engine's files form one
section named after it. Board pages drop the heading that repeated their
title, parts drop the Part of line their crumbs already say, and pages
that moved redirect.
A file most of whose readers take beside another one is no start file
(NAOMI opened on hod2bios.zip), and names that answer one slot for
different emulators read as alternatives with their readers, each by
the name of its card.
The API reads release_pack from the registry entry, not the inherited config, and the picker gets release-advertised.json. The lookup counts emulators that read a file by name, every record names what was declared, the Data page lists all schemas, the target lists and nested manifest fields, and systems.json items carry their api address.
A profile pinned to one platform's build (pin_frozen) answers after the emulator, as a frozen_snapshot does: Flycast now answers dc bios missing ahead of Reicast and Flycast 2021 LE. terms weighs 30, so the version table answers which mame version first, and the gate now asks for rank 1 there. The search field test names the five fields, since round 4 item 2.4 put the title meta on the h1.
The glance on the installer page is written from install.py's argument parser when the page is built, grouped by GLANCE_GROUPS. An option no group names lands under Other, so a new option cannot be left off the page.
Root pages nothing produced are swept, so a table moved to a platform page no longer keeps the address its redirect needs. Data tree pages name their readers, board crumbs end on the Arcade page once, fba2012 is headed FB Alpha 2012, test profiles get no page, and the installer glance lists --declared-only and --offline. sitesources.py goes through run_cli.
check_freshness reports source-paths.json STALE when it records a tree no profile pins any more, and tools.md names sitesources.py --fetch where pins move.
BizHawk's tile takes EmuHawk's application icon (ApplicationIcon in its csproj) instead of an account avatar. MinUI has no logo of its own: its entry and copy go, and fetch_site_images.py now removes the copy of a platform without a logo.
validate_schemas.py read install-events.schema.json from the real tree while a test pointed ROOT at a scratch one, and test_schema_semantic_gate.py called unittest.main() before its last class.
Abdessamad Derraz